A research team has made a significant breakthrough by creating molecules that have the potential to change the landscape of computing.
A team from the University of Limerick has achieved a remarkable discovery by crafting molecules that may transform the field of computing.
The scientists at UL’s Bernal Institute have identified new methods for examining, managing, and customizing materials at the fundamental molecular level.
The findings have contributed to an international initiative involving global experts aiming to develop a revolutionary hardware platform for artificial intelligence, resulting in extraordinary enhancements in computational speed and energy efficiency.
This research has recently been published in the prestigious scientific journal Nature.
The UL team, under the guidance of Damien Thompson, a Professor of Molecular Modelling at UL and director of SSPC, Ireland’s Research Centre for Pharmaceuticals, collaborated with researchers from the Indian Institute of Science (IISc) and Texas A&M University. They believe that their findings will pave the way for innovative solutions to significant societal challenges related to health, energy, and the environment.
Professor Thompson stated: “Our design takes cues from the human brain, employing the natural movement of atoms to process and store data. As the molecules rotate and vibrate within their crystal structure, they generate numerous distinct memory states.
“We can chart the movement of the molecules inside the device and link each moment to a specific electrical state. This generates a sort of travelogue for the molecule, which can be written and accessed in a manner similar to conventional silicon-based computers, but here with vastly improved energy efficiency and space utilization, as each entry is smaller than an atom.
“This unconventional solution holds great promise for various computing applications, from power-hungry data centers to memory-intensive digital mapping and online gaming.”
To this point, neuromorphic platforms—computing systems inspired by the human brain—have been limited to low-accuracy tasks, such as inference in artificial neural networks. This limitation exists because essential computing functions like signal processing, neural network training, and natural language processing demand significantly higher computational precision than current neuromorphic circuits can deliver.
As a result, achieving high precision has posed the most formidable challenge in neuromorphic computing.
The team’s reimagining of the foundational computing architecture meets the demand for high precision, executing resource-intensive tasks with remarkable energy efficiency of 4.1 tera-operations per second per watt (TOPS/W).
This advancement propels neuromorphic computing beyond its traditional niche applications, potentially unlocking the long-anticipated transformative advantages of artificial intelligence and enhancing the core functionality of digital electronics, from cloud systems to edge devices.
Professor Sreetosh Goswami, the project lead at IISc, remarked: “By meticulously managing the extensive range of molecular kinetic states, we have developed a highly accurate, 14-bit fully functional neuromorphic accelerator that integrates seamlessly into a circuit board capable of handling signal processing, AI, and machine learning tasks, including artificial neural networks, auto-encoders, and generative adversarial networks.
“Importantly, taking advantage of the accelerators’ high precision allows us to train neural networks at the edge, tackling one of the most urgent challenges in AI hardware.”
Further advancements are on the horizon, as the team continues to explore a broader range of materials and methodologies to enhance the power of these platforms even more.
Professor Thompson added: “Our ultimate goal is to replace conventional computers with high-performance ‘everyware’ constructed from energy-efficient and eco-friendly materials that enable distributed information processing integrated into everyday objects, from clothing and food packaging to building materials.”
